OpenAI Drops Ten Math Results. Mathematicians Cry Foul. Peer Review Is Not Optional, Clearly.
OpenAI announced 10 AI-generated mathematical advances found during internal testing of its next major model. Francesco Fournier-Facio, a group theorist at Cambridge, initially found the results stunning. Upon applying standard academic scrutiny, he and colleagues concluded the work fell short of accepted research rigor.
This illustrates the principle of epistemic asymmetry. A system can produce correct-looking output without following the validation norms that make knowledge trustworthy. The mechanism is what I call unearned authority: when provenance confers credibility that methodology has not earned. Always audit the process, not just the result.
OpenAI released the results. Francesco Fournier-Facio at the University of Cambridge and colleagues performed the critical review. The verdict from experts was that standard academic rigor was absent.
- Open any chatbot and ask it to prove a simple mathematical statement, such as why the square root of two is irrational. Read the full response carefully.
- Search for a human-written proof of the same statement using a university lecture note or textbook. Compare the logical steps and assumptions.
- Identify one claim in the AI proof that lacks justification or skips a logical step. Write down what is missing. You have now performed basic peer review.